Local Structural Features Elucidate Crystallization of Complex Structures.

Area of Science:

• Materials Science
• Computational Chemistry
• Soft Matter Physics

Background:

• Complex crystal structures exhibit diverse local environments, but their spontaneous emergence during crystal growth remains poorly understood.
• Understanding the self-assembly of ordered structures from disordered states is crucial for materials design.

Purpose of the Study:

• To investigate the emergence of local order during crystal growth across various structures and pathways.
• To develop and apply a machine learning method for identifying and classifying local atomic environments during crystallization.

Main Methods:

• Utilized self-assembly simulations of identical particles interacting via multiwell isotropic pair potentials.
• Applied an unsupervised machine learning method to bond-orientational order metrics to identify local motifs.
• Analyzed crystallization pathways from fluid to amorphous liquid droplets to bulk crystals.

Main Results:

• Successfully distinguished different crystallographic sites in complex structures using the machine learning approach.
• Observed consistent under-coordination in the liquid phase compared to the average coordination number in the bulk crystal.
• Analyzed the geometrically frustrated growth of a Frank-Kasper phase, identifying competition between defects and high-coordinated sites.

Conclusions:

• The particle-level classification method provides a powerful tool for studying structural self-assembly and crystal growth.
• Findings offer insights into the formation pathways of complex soft-matter structures and can guide the design of new building blocks.